¹ Typical examples may be seen on the ingots illustrated in JRS XXI (1931), 260, fig. 28, pl. XXIX and XXX.

² J. Brit. Arch. Assoc. XVI (1860), 330; Intell. Obsvr. XI (1867), 354; cf. Haverfield, in VCH Shropsh. I (1908), 264Google Scholar, no. 1.

³ So Watson, R., Chem. Essays (4th ed., 1787) 111Google Scholar, 267, of the Cromford Moor ingot (CIL VII, 1208: Brit. Museum).

⁴ Way, A., Arch. J. XVI (1859), 35Google Scholar (Matlock Moor ingot, CIL VII, 1214: Brit. Museum).

⁵ Gent. Mag. LX (1790), 525 (Shropshire ingot, CIL VII, 1209a); Pegge, S., Archaeologia IX (1789), 46Google Scholar (lost Matlock Bank ingot, CIL VII, 1215a).

⁶ Archaeologia LVII (1901), 398 (cf. Proc. Soc. Antiqs. ² XXXI (1918), 39, in his discussion of the Bitterne ingots).

⁷ See details in JRS XLIVII (1957), 230–1, no. 20.

⁸ By ProfessorPalmer, L. S. and MrAshworth, H. W. W., in Proc. Som. Arch. and NH Soc. CI/CII (1956–1957), 52–88.Google Scholar

⁹ Trans. Newcomen Soc. XX (1939–40), 139.

¹⁰ Between 1930 and 1933 I had been able to examine most of the ingots in this country; soon afterwards, in collaboration with Dr. Smythe, a re-examination of all surviving British specimens was made with special attention to metallurgical details. Later, having noticed that surface markings apparently similar to those on Roman ingots appeared on blocks of linotype metal which were being cast in one operation, I suggested that some trial casting might be useful; this was undertaken by Dr. Smythe between 1937 and 1939 and I was able to observe the whole series of experiments, so correcting my own earlier acceptance of the supposed lamination of ingots (Trans. Newcomen Soc. XII (1931–2), 62.)

¹¹ Moulds used were variously made from sand, copper, marble, heat-insulating brick and fireclay, either dried or baked (at c. 800° to 900° C.). This last gave the most satisfactory results comparable with the Roman ingots, and it is these results that are discussed above.

¹² For brevity in dealing with a somewhat complex matter (a) reference is made above only to the sides and ends, and not to the bottom face (as cast) of the ingots; (b) no regard is paid to the many variables which in practice can affect the casting temperature—and hence the surface appearance either of the whole or of various parts of the ingot, e.g. atmospheric or local conditions, temperature of the mould above that of the atmosphere, i.e. mould already warm or hot from previous use, variation of pouring spot in relation to the mould, or—a most important factor—variation in the rate of pouring.

¹³ The truth of the matter could indeed have been grasped long since, if proper attention had been given to some of the ingots. The one found at South Cave, e.g., (Eph. Eplgr. IX, 1265; Hull Museum) has suffered damage which clearly shows its inner part to be a solid mass. But admittedly striations, though present, are not very obvious on this ingot.

¹⁴ Not of course merely the initial temperature and fluidity of the casting metal, but the actual temperature which it maintains on reaching any particular part of the mould. This can be affected in a variety of ways (cf. n. 12 (b) above), even during a single casting operation.

¹⁵ So also in very similar words Palmer and Ashworth (o.c. n. 8 above), 71.

¹⁶ CIL VII, 1214 (cf. n. 4 above).

¹⁷ Since Gowland's study of the ingots was largely concentrated on those preserved in the British Museum it is likely that the Verecundus ingot, in conjunction with his misunderstanding of the nature of the striations, convinced him that all ingots were cast piecemeal. His sketch of this ingot (o.c. (n. 6), pl. LVII no. 2) gives some slight indication of one of the three fissures that mark off the four layers; the ‘flats’ are fairly accurately defined.

¹⁸ Arch. J. LII (1895), 33; Haverfield, , Viet. Co. Hist. Derbys. I (1905), 232Google Scholar, no. 13. Now in Weston Park Museum, Sheffield. Its Roman date, though probable, cannot be beyond all doubt. Clear striations on parts of its surface prove that it was cast in a mould, not ‘run into a basin of sand’, as once asserted (Antiquary XXX (1894), 46). The mould, however, must have been warped and damaged as well as roughly made.

¹⁹ J. P. Bushe-Fox, First Report Richborough (Soc. Antiq. Research Rep. (1926) 42 and pl. XI). The small separate layer is not there shown, and no reliance can be placed upon the markings apparently suggesting striations.

²⁰ Eph. Epigr. III, 121 (e); IV, p. 206; Haverfield, , Vict. Co. Hist. I (1906), 343Google Scholar, nos. 14 and 15. Both now in Taunton Castle Museum. As Haverfield suspected, they are separate fragments and not parts of a single cast; examination shows that they were produced by molten metal which entered the mould from a different direction in each case. Both were originally more extensive in area, but it is not possible to say whether either was large enough to have reproduced the full inscription (for which cf. CIL VII, 1211, the lost Bruton ingot).

²¹ (a) CIL VII, 1217. Found at Brough-on-Humber early in the eighteenth century, and recorded as inscribed BREXARC (i.e. BR·EX·ARG; cf. the latter part of the inscription on other ingots since found in the same area: cf. n. 9 above); size 3½inches (a normal width) by 7 inches (rather more than one-third of the normal length of a full ingot), thickness not stated. Horsley, Brit. Rom. (1732) 192, N.65 (Yorkshire), XIX, and p. 314, XIX; Haverfield, Viet. Co. Hist. Derbys, 1 (1905), 231, no. 5; Gowland, 1.c. 402–3, n. (a). (b) Eph. Epigr. 111, 121 (c); Haverfield, Vict. Co. Hist. Som. 1 (1906), 341–2, no. 8. Inscribed IMP VESPASIA and measuring 3½ inches by 15 inches by 2 inches thick. It was thus (like (a) above) of normal width but not full length. A depth of 2 inches would mean a full length ingot of not more than approximately 66 lb. weight, which is too low for a completed casting, especially as all others of Vespasianic date known from the Mendips weigh nearly three times this amount, as also do those from Flintshire, except the pair found on Hints Common, Staffs, of 151 lb. and 150 lb. (CIL VIL, 1205: Brit. Museum; Eph. Epigr. IX, 1264: Tamworth Castle Museum).

²² The statement that ‘bar-ingots were cast from several crucible-charges’ (Davies, O., R. Mines in Europe (1935) 149Google Scholar) is of course equally unacceptable in view of what is said above.

²³ The specific gravity of solid lead is approximately 11·3, but that of molten lead approximately 10·8.

²⁴ Only a few instances can be given here. Contraction furrows have often been filled up with makeweight metal; but good examples can be found, e.g. on the two Domitianic ingots from Yorkshire (CIL VII, 1207: Brit. Museum and Ripley Castle). External pipes, single or double, of varying size have been found on over twenty ingots, most of them only discoverable by probing in the first place owing to soil and corrosion crust, some being concealed under partially detachable make-weights. Existence of internal pipes is naturally difficult to confirm. Fortunately two fragmentary ingots survive which show at least portions of such pipes; one, in the Richborough ingot (cf. n. 19 above), is round with a diameter of between ½-inch and 1 inch, and it extends for about 4 inches into the body of the ingot (pl. VI); the other, in the end portion of an ingot found at Caerwent in 1947 (JRS XXXVIII (1948), 81 and 101, no. 7), is roughly rectangular (approximately ⅜-inch by ⅝-inch) at the mouth, and extends for about 2½ inches, becoming funnel-shaped. Drilling (for specimens for silver assay) has also accidentally revealed three such pipes; one of these, in the halfingot of Severan date found at Lillebonne but exported from Britain (CIL XIII, 3222: Musée Dép. Rouen) lay 1½ inches below the open top surface of the ingot as cast, was approximately 1½ inches in diameter and extended for nearly 7 inches longitudinally. It is worthy of note that an ingot of the Spanish semi-cylindrical type found in the Tiber (CIL XV, 7917: Antiq. Comunale, Rome) originally complete, has been sawn asunder so as to preserve only a strip about 1½ inches thick carrying the inscription. This has revealed, on the sawn surface, the remains of an internal pipe, roughly pear-shaped, approximately 1 inch maximum width and about 2 inches in length. Such internal pipes may well exist in a number of ingots, and would help to explain any notable discrepancies between calculated volumetric and actual found weights. In the experimental ingots the position of both external and internal pipes could be made to vary in relation to varying conditions of casting.

²⁵ See n. 9 above, and cf. JRS XXXI (1941), 145–6, nos. 15–16, pl. XX.

²⁶ Total length approximately 31 inches: in outline slightly wasp-waisted, varying from about 8 inches maximum width at the more shallow end (marking the earlier flow of hotter and more fluid metal) to 3½ inches in the waist; then approximately 4 inches for the remainder, where the depth is almost double that of the shallow end due to the decreasing head of metal and the general drop of temperature as the rate of flow declines.

²⁷ Possibly, e.g. no mould was available (there is strong evidence that they were not numerous) because a previous casting was not yet cool enough to tip out; or an open pour in this manner may have been preferred because the amount of molten metal available was known to be greatly deficient for casting an ingot of the approximate weight of 190 lb., which was evidently the standard for this area at this period.

One is reminded of the mass found in May, 1875, at Charterhouse on Mendip (H. M. Scarth's note of January, 1876, in Proc. Bath N. Hist, and Antiq. Field Club 111, 340; cf. Proc. Soc. Antiqs. ² VII (1877), 156–7), and Haverfield, in Vict. Co. Hist. Som. I (1906), 337)Google Scholar, described as ‘a cake of pure lead weighing 78 lb. which had been left just in the condition in which it had run out of the ore into the hearth, the under surface is quite flat, the upper surface is slightly rounded’. The explanation of this as a bun-ingot (O. Davies, o.c. (n. 22) 149) is improbable, as there is no evidence for such in Roman lead working in Britain; if it is Roman it is much more likely to have been an open casting similar to that from Brough-on-Humber, but made on to a more level floor. The mass is lost and no further details are known.

²⁸ Eph. Epigr. III, 121 (d): see my notes in JRS XXI (1931), 259 ff., with fig. 28 and pl. XXX.

²⁹ CIL XV, 7915 (National Mus., Rome). The weight is not given in CIL, nor by Lanciani, (Acque e Acqued., Rome (1880), 191)Google Scholar, but Dr. Paribeni in 1953 kindly confirmed it as 275 kg., substantially that given by Besnier (Rev. Arch.⁵ XIV (1921), 114, no. 69).

³⁰ Now in the Museum at Hälsingborg. I have not had an opportunity to examine the ingot, but from details and photographs generously sent by Dr. F. Mårtensson, the Director in 1938, it appears to have been cast at a single pour but in a mould already containing a thin sheet of lead from a previous casting; the mould used was less regular in shape and with rougher inner faces than was that of the Tiber ingot. Cf. Täckholm, U., Stud. ü. d. Bergbau der röm. Kaiserzeit, Uppsala (1937), 172–3Google Scholar, where a Hadrianic date is suggested; but operations in the Ralja area may well be at least a quarter of a century later (cf. Davies, O. in JRS XXVII (1937), 284).Google Scholar

³¹ See O. Davies, o.c. (n. 22) 43 ff. At Pentre furnaces with at least foundations of solid masonry were found in 1923 (Atkinson, and Taylor, , Flintsh. Hist. Soc. Publ. X, i (1924), 7 ff.)Google Scholar, in association with a pit which supplied clay for the structure (l.c. 10) and lead fragments apparently from channels by which the molten lead ran into the moulds. Finds in 1924 (o.c. X, 2 (1925), 8) included a mass of slag ‘which from its shape showed that it had formed round the interior of a furnace tap-hole’, and ‘indicated that the diameter of the tap-hole had been about 2 inches’.

³² Casting in the same mould can be established not only by dimensions and angles but more positively by precise comparison of flaws in the lettering or other details, due to damage sustained by the mould in course of constant use. This method proves the existence of seven pairs, one group of four and one of five. Other groups of three and four, recorded as found together but now lost, are likely to have been similarly the products from a single mould in each case.

³³ I should like to express sincere thanks to Messrs. Associated Lead Manufacturers, Ltd., and in particular to Mr. J. O. Hudson, Manager, and Dr. R. Bell, Assistant Manager, of the Elswick Works, Newcastle upon Tyne, not only for their interest and help in my inquiries during 1959 but also for taking and allowing me use of a series of photographs from which those in pl. VII, 2, 3 have been selected.